Indications and implications of a borehole seismic monitoring of the San Andreas fault envision a viable way for earthquake prediction

The Parkfield earthquake experiment not only failed to predict the 2004 M6 earthquake but has not found satisfactory answers to why this earthquake was mis-predicted nor clues on how to predict future earthquakes. Seven months after the earthquake, we deployed a seismic array in the SAFOD main hole next to the SAF, to monitor the fault for 13 days. By analyzing the projected hypocenters of the best 100 earthquake events, we found that the SAF at the SAFOD site is composed of three indispensable active sub-faults, namely SAF1, SAF2, SAF3. During the monitoring period, it seemed that the move of SAF1 drags SAF2, which in turn drags SAF3 to move one after another, not just in the SAF's strike direction, but at an approximate 45 degrees dipping angle. The result also hints the existence of relatively strong asperities in the northwest direction on SAF1. However, our monitoring suggests that at the SAFOD site during the 2004 M6 earthquake, the main rupture took place on SAF2, which could also extend with a bend near Parkfield town to the individual Southwest Fracture Zone, and to the seemingly isolated hypocenter of the 2004 M6 mainshock. The possible prevailing of this triple active sub-faults model for the SAF in the Parkfield area may account for the recurring timing of M6 earthquakes which depends on how many of these three active sub-faults have basically absorbed the tectonic stress, which can result in a cycle of approximately either 12, 24 or 36 years, instead of a single 22 years as previously believed that led to the USGS's 11-years error in predicting the 2004 M6 earthquake. In a mere 0.05% of the earthquake's last intermission time, and at just one location, our borehole seismic monitoring has revealed so much insightful information about the SAF's complex structures and seismic spawning and developing dynamics, that we believe earthquake prediction can be feasible.